X's algorithm shifted users toward conservative views, study finds

Algorithmic feeds shift users rightward

Switching users on X from a chronological feed to the platform's algorithm shifted their political opinions toward more conservative positions, according to a new study. The 2023 field experiment found the change affected views on policy priorities, investigations into Donald Trump, and the war in Ukraine.

Researchers randomly assigned active, US-based X users to either an algorithmic or a chronological feed for seven weeks. They then measured changes in political attitudes and online behavior. The shift toward the algorithm increased user engagement and produced the conservative shift.

Effects are one-way and persistent

Critically, the effect only worked in one direction. Switching users from the algorithmic feed to the chronological one did not produce a comparable liberal shift in opinions.

The study found that neither turning the algorithm on nor off significantly affected affective polarization or a user's self-reported partisan identity. The changes were in specific policy attitudes, not overall tribal affiliation.

Algorithm promotes conservative content

To understand the mechanism, researchers analyzed the content of users' feeds. They found X's algorithm systematically promotes conservative-leaning posts and demotes content from traditional media outlets.

This exposure had a lasting behavioral impact. Users assigned to the algorithmic feed began following more conservative political activist accounts. They continued to follow these accounts even after the experiment ended and they switched back to a chronological feed.

This "follow" behavior helps explain the asymmetry. The algorithm introduces users to new, right-leaning sources they then choose to keep in their network, creating a persistent effect.

Contradicts earlier Meta findings

The results present a puzzle compared to earlier research on Meta platforms. A previous large-scale experiment that turned off Meta's feed algorithm found no significant effects on political attitudes.

The X study suggests a platform's specific algorithmic design is crucial. X's recommendation system appears uniquely effective at altering user networks and, by extension, their political views. The key factors include:

• Promoting conservative activist content over traditional media
• Incentivizing users to follow new right-leaning accounts
• Creating lasting network changes that survive the algorithm's removal

Algorithms shape networks, not just feeds

The core finding is that initial exposure to X's algorithm has persistent effects. It changes whom users follow, which in turn shapes the content they see and the opinions they form over time.

This network effect persists even when the algorithm itself is switched off. A user's curated following list continues to deliver a skewed information diet.

The study concludes that while feed algorithms may not change your stated party affiliation, they can significantly and lastingly alter your policy views by reshaping your information network. On X, that shift consistently pulls users to the right.